tecosystems

R&D is Important, But For Whom?

Share via Twitter Share via Facebook Share via Linkedin Share via Reddit

It takes a kind of institutional patience — because the payoff from fundamental research usually doesn’t come in a quarter…or a year…or even sometimes in a decade — if it ever comes.

Not only that, but there are constant pressures on this kind of commitment. Shareholder expectations for higher returns don’t diminish when the economy stutters.

In these moments, the temptation is strongest to cut investments in skills and R&D. We’ve seen more than one example of that here in the Valley in recent years.

And yet, Tom Watson Sr. actually increased research investment during the Great Depression. And his son bet the business on the System/360… a bold move chronicled so powerfully here at the museum.” – Sam Palmisano, CEO IBM

The question of the importance of research and development is an interesting one. In the abstract, R&D seems important, even vital. In practice, however, its results are frequently uneven and often non-existent.

If we were to objectively look at the technology industry’s opinion of the importance of R&D, what might we look at? One metric might be R&D spend a function of revenue. As Asymco’s Horace Dediu pointed out, “Samsung spends 6.5% of sales on R&D. Apple spends 2.2%.” But how does that compare to the industry. Is Samsung high? Apple low? Or neither.

To explore this question we compiled the R&D and revenue numbers for an arbitrary collection of technology vendors for a five year period from 2005-2010. From these figures we prepared the time series motion chart (user guide [PDF]) below.

[cf]RD[/cf]

This graph addresses some of the questions above. Apple’s R&D spend is consistently among the lowest in our sample, for example, eclipsed periodically only by HP. But it generates more questions than it answers.

Example: if IBM prioritizes R&D, why are they consistently below the median five year spend of 12.8%? Conversely, why are VMware’s R&D costs so high, on a relative basis?

In general, however, the variation in R&D as a percentage of income is wide, ranging from 2.35% to 24.53%. Which is interesting. What’s more interesting is that there does not appear to be an obvious connection to market performance. If you compare the growth in mean share prices for the five year span to average R&D spend, there is no apparent correlation.

None of which is to say that R&D isn’t important; it is self-evident that the reverse is true. Much of what we take for granted today – from the processor to the LCD display to the mouse driven user interface – is the product of research and development. From a user’s standpoint, then, R&D is vital.

What the above suggests, however, is that R&D is relatively independent of financial performance for companies that conduct it, with the caveats that this is merely a five year sample and that the selection of vendors is arbitrary and cross-industry. Which in turn implies that Palmisano is correct, and that we should consider research and development separately, because the two may actually have little to do with one another.

Anyhow, explore the motion chart and see what patterns you can detect. Try pressing play, for example, and watching the R&D spend fluctuate over the last five years.

Disclosure: IBM and VMware are RedMonk customers, Apple is not.

19 comments

  1. This may be a case to also look at it from the perspective of absolute R&D spend and a companies ability to grow above the average at a certain investment level. While as a % of revenue may be a common metric, if you think about the drive to invest in R&D it’s so that you can develop a product or set of products that are high value and drive revenues faster than expenses. Case in point, Apple -> If they had to spend a fixed % on R&D as they grew, it would be less powerful of an investment than if they could spend a lower $X on R&D and grow it as much as the market would continue to accept/buy it. The % model is at odds with what companies try to do as they invest in R&D – exploit that innovation as much as possible.

    1. The difference between fixed and proportional R&D seems to me to be whether the company wants to expand its offerings or replace them. Apple might be a good example for fixed-rate R&D, since it’s primarily coming out with replacements for things that already exist or a very small number of new products. But companies growing their markets rather than selling more of a single product would probably be more interested in proportional R&D.

  2. I also think you need to look at absolute spend — the range in revenues is broad for the companies you are comparing. A quick trip to Wolfram Alpha gave me (2010?) revenues of > $100B for HP, IBM, Apple (lowest % spend) down to a lowly $1B & $3.3B for Red Hat and VMware (highest % spend). Comparisons within each group (S, M, L) might be interesting, but I’m guessing that looking at the composition of businesses within each company (IBM’s broad portfolio of hw, sw, and consulting vs Apple, for instance) would explain many of the differences. Apple ends up being the freakish outlier from many perspectives.

  3. @Mike Dolan: The percentage model is imperfect as well because it discounts margin, i.e. some inventions are higher margin than others. So it has little to say about efficiency of innovation.

    It is informative, however, as a rough indication of big picture costs. How much do respective entities have to spend to maintain their position? Is it harder for them, on a relative basis, or easier?

    No one should expect to spend a fixed percentage on R&D, because that’s not reflective of need. But if you’re investing substantially more or less than competition, it’s probably worth asking why.

  4. I would imagine that there is also a very significant difference in how costs are allocated to R&D in each of the companies balance sheets. I.e. two companies may carry out superficially similar activities, but the cost of doing those activities may well get accounted for in different ways. Comparing the absolute values of items in financial statements (especially intangible ones like R&D) is fraught with difficulty.

    1. I’m sure there are discrepancies in terms of R&D reporting; one example is companies that like to showcase their spend, intending it as a proxy for their innovation pipeline. 

      But accounting specifics aside, the relative investment levels – or, as you might argue, the “reported” investment levels – are interesting in context. 

      Or so I think, anyway.

  5. I would imagine that there is also a very significant difference in how costs are allocated to R&D in each of the companies balance sheets. I.e. two companies may carry out superficially similar activities, but the cost of doing those activities may well get accounted for in different ways. Comparing the absolute values of items in financial statements (especially intangible ones like R&D) is fraught with difficulty.

  6. I would imagine that there is also a very significant difference in how costs are allocated to R&D in each of the companies balance sheets. I.e. two companies may carry out superficially similar activities, but the cost of doing those activities may well get accounted for in different ways. Comparing the absolute values of items in financial statements (especially intangible ones like R&D) is fraught with difficulty.

  7. BTW, the motion charts are awesome – very impressive – how long did that take to create? 

    1. @Mike Dolan: thanks! thanks to the Google Motion Charts w/ R library (http://code.google.com/p/google-motion-charts-with-r/), it’s pretty easy. if you use R, of course. 

  8. IMO the real question is not how much you spend but how well integrated the R&D efforts are to the rest of the business. 

    In some spaces (say material science for chip makers, nano-scale manufacturing etc.) there’s tremendous lead time between research and production use – those areas also require huge capital investment to do real research. In those areas having a significant separation between R&D and product/manufacturing is unavoidable.In software the picture is different: you cannot separate R&D from real product development too much or else you’ll create ivory towers. I spent 8 years in IBM Research and unfortunately (IMO; I’m sure others will argue otherwise) much of the software research there lacks the deep integration to the “real world” to end up producing the next Facebook or whatever. There are of course deep long term software problems that need to be solved, but IMO its best for the world for those to be done in more open settings such as national labs or university research facilities to enable broad adoption (I’m thinking patents here). So whether you spend 1% or 10% what matters most is how directly and iteratively coupled the R&D teams are with the teams that make stuff that the organization sells. Apple seems to be spending less on R&D but no one will of course question the amazingness of their innovation engine. This is how WSO2 is designed .. R&D is an integral part of our product teams and broader technology teams. Someone who does R&D does not get to “throw it over the wall” and move onto more R&D – they need to make a product out of it (or integrate it to some existing product), evangelize it, sell it to customers, help customers use it, improve it etc.. Essentially we don’t have anyone in R&D more only and we never will. IMO having an R&D team that is disconnected, differently incentivized etc. is the reason for ineffective R&D. For longer term R&D stuff we work with academia to enable, encourage and guide them.Of course we’re a puny little company compared to the guys you’re looking at :-).

    1. There’s little question that closer examinations of R&D necessarily need to deconstruct the nature of the research as well as examining critically the developmental components. R&D is a process, and in some contexts is best embedded in with commercial processes, as you state. 

      The difficulty, however, is that research that is tightly coupled with commercial interests is likely to miss new and disruptive opportunities, because the existing businesses are unable to perceive them sufficiently. 

  9. […] spends 6.5% of sales on R&D. Apple spends 2.2%,” points out this analyst, but Apple keeps trying to block this Linux-based competition from Korea.█ Share this post: […]

  10. Stephen, on your last point, there is very little commercial research activity that is not tied to product strategies. IBM perhaps come closest today but even that is probably more coupled to IBM’s (wide-ranging) interests than in the past.

    I think the Apple number is something of an outlier in large part because Apple is so bloody successful. They sell so much high margin stuff within a relatively narrow scope of products that there’s just a huge multiplier effect. It’s a little like manufacturing investments in the old days for companies that were running flat out vs. those that had a lot of idle capacity–but more so.

  11. @Mike Dolan: The difference between fixed and proportional R&D seems to me to be whether the company wants to expand its offerings or replace them. Apple might be a good example for fixed-rate R&D, since it’s primarily coming out with replacements for things that already exist or a very small number of new products. But companies growing their markets rather than selling more of a single product would probably be more interested in proportional R&D.

  12. […] La ricerca è importante. Ma per chi? In questi momenti di crisi la tentazione di fare come la Banda del Nano Parruccato è forte. Apple spende pochissimo, a differenza di altri. ::: Redmonk […]

  13. I would envision that there is also a very important differentiation
    in how expenditure is allocated to R&D in each of the company’s equilibrium
    expanses.
     

  14. Pretty hopeful discussion about the importance of R & D. I would like to allocate this topic with others. Thanks mate 🙂

Leave a Reply

Your email address will not be published. Required fields are marked *